ZenAI DigitalContact
Home / Assessments & Tools / Impact Screening
Assessments & Tools

AI Use-Case Impact Screening

Understand who your AI initiative could affect, identify potential harms and see which safeguards need attention.

This screening supports an initial discussion of one use case. It does not assess organisation-wide maturity.

Answer for a specific AI initiative, based on what you know today. Use Unsure where further investigation is needed. No answers are preselected.

Do not enter personal information or confidential project details. Screening responses stay in this page and are not submitted to ZenAI. Privacy notice

How this screening works and references

Use-case context (optional)

1. Potential impacts

A Yes identifies an area for examination. It does not automatically mean the use case is unacceptable.

1 Could an incorrect output materially affect someone's access to a service, opportunity or entitlement?
2 Could failure or misuse cause physical harm, serious distress or substantial financial loss?
3 Does the use case process personal, sensitive or confidential information?
4 Could it disadvantage particular groups or people who cannot easily challenge its outputs?
5 Can the system communicate externally, change records or initiate transactions without prior human approval?

2. Safeguard preparation

Select the status you can support. Evidence remains self-reported in this screening.

6 Have you defined the intended use, its boundaries and the outcome you want to achieve?

Evidence to consider: Use-case brief and success criteria.

7 Have you identified affected people and considered their perspectives?

Evidence to consider: Stakeholder review or consultation.

8 Have you checked whether the data and supplier arrangements suit this use?

Evidence to consider: Data review and supplier documentation.

9 Have you tested plausible errors, misuse and different user circumstances?

Evidence to consider: Test cases and findings.

10 Have you assigned authority to review, challenge, override and stop the system?

Evidence to consider: Responsibilities and intervention procedure.

11 Have you agreed how to monitor impacts, handle incidents and reassess changes?

Evidence to consider: Monitoring and review plan.

0 of 11 questions answered

Method, interpretation and references

How this lightweight screening was developed

ZenAI Digital selected topics for an initial discussion of AI use-case impacts and safeguard preparation. This screening uses original ZenAI questions and interpretation rules. It is not a validated shortened version of an external instrument.

Interpretation and weightage

  • There is no weighted risk score or overall pass/fail result.
  • Each Yes in part 1 produces an impact area and a suggested review action. Each Unsure remains an unresolved question. A No does not establish absence of risk.
  • Each of the six safeguard areas counts equally towards reported preparation coverage only when selected as established with supporting evidence.
  • Partial, missing and unsure safeguards remain visible separately. Safeguard answers cannot cancel potential impacts.
  • These are proposed ZenAI screening rules, not empirically validated risk weights, regulatory classifications or predictions of success.

References and topic mapping

Microsoft Responsible AI Impact Assessment Template (June 2022)
Reference topics: system context (section 1), stakeholders and human oversight (section 2), adverse impacts (section 3), data (section 4) and impact review (section 5). Linked for reference. Its complete questionnaire, Microsoft-specific requirements and approval process are not reproduced.
Australian Government DTA AI Impact Assessment Tool and guidance
Australian public-sector context and a route to fuller use-case assessment. Consult the applicable agency process. This screening does not establish satisfaction of DTA or agency requirements.

Scope and limitations

Method version 0.1. This preliminary screening does not cover every hazard, sector requirement or jurisdiction. No third-party endorsement is implied. The references support topic selection, not validation of these questions or rules. Comprehensive work requires evidence review, stakeholder input and appropriate specialist judgement.

Revision: 27 October 2026. Initial screening design. Your result and the referenced resources can support a more detailed discussion.